Journal of Applied Behavior Analysis 2020, 9999, 1–19 NUMBER 9999 ()

An evaluation of matrix training to teach college students piano

notes and rhythms
Emily K. Langton, Caio F. Miguel, Jocelyn E. Diaz, Maria Clara Cordeiro
and Megan R. Heinicke
California State University, Sacramento

Twelve college students learned to tact the names of notes and rhythms and play them when
presented with compound stimuli (visuals of notes and rhythms on a musical staff). In Experi-
ment 1, we assessed generalization by presenting novel notes, rhythms, and compound stimuli
not previously paired together. In the second experiment, we added a metronome that played at
60 beats per minute in all conditions for 3 out of 6 participants to ensure consistent tempo.
Across both experiments, participants passed almost all posttests with the exception of tacting
and playing in the presence of sound clips. Our data suggest that matrix training is an effective
procedure to teach music skills to college students.
Key words: matrix training, music, notes and rhythms, recombinative generalization

Music education supplements the overall typically developing children and those with
intellectual development of children (Hudziak developmental disabilities (Griffith et al., 2018;
et al., 2014; Lim, 2010; Lim & Draper, 2011) Hill et al., 2020). Research suggests that leisure
as evidenced through improvements in lan- skills may reduce stress and depression, cogni-
guage acquisition (Brandt et al., 2012), math tive aging, and improve memory (Hanna-
(Graziano et al., 1999; Vaughn, 2000), reading Pladdy & MacKay, 2011; Mannell, 2007) in
(Zuk et al., 2013), social interactions and com- young adults and elderly individuals. Music
munication (Brown & Jellison, 2012; Gourgey, educators have used various methods to teach
1998; Sussman, 2009). Music training may individuals basic musical and instrumental skills
also support the development of children with regardless of age or ability (Farber & Parker,
developmental disabilities as evidenced by 1987; McRae, 1982; Russell-Smith, 1967;
increased IQ scores (Kaviani et al., 2014; Suzuki Association of the Americas Inc., 2011),
Schellenberg, 2006), improvements in motor and both students and teachers anecdotally
skills (Gilbert, 1983) and joint attention (Kim report success with different procedures. How-
et al., 2008), as well as decreases in vocal ste- ever, to our knowledge, there are no empirical
reotypy (Rapp, 2007). In addition, music may studies aimed at evaluating the effectiveness of
serve as an appropriate leisure skill for both these methods.
Within music education, there is a system to
This study is based on a thesis submitted by the first understanding and reading music. There are
author under the supervision of the second author to the
Department of Psychology at California State University, seven letters (i.e., A, B, C, D, E, F, and G) that
Sacramento in partial fulfillment of the requirements for are called music notes and represent a specific
an M.A. degree in Psychology: Applied Behavior Analysis. pitch (Schonbrun, 2005). Pitch simply refers to
We would like to thank Svea Love for her assistance with
data collection and Judah Axe for his comments on a pre-
the sound a note can produce. Pitches may
vious version of this manuscript. repeat at higher and lower intervals. The music
Address correspondence to: Caio F. Miguel, Depart- staff is a visual system used to read and write
ment of Psychology, California State University, Sacra- notes composed of five lines and four spaces,
mento, 6000 J Street, Sacramento, CA 95819-6007.
Email: [email protected] each of which represents a specific pitch. The
doi: 10.1002/jaba.690 higher a line is, the higher the pitch and vice
© 2020 Society for the Experimental Analysis of Behavior
1
2 Emily K. Langton et al.

versa. Another important aspect of reading of chords that made up a musical piece on the
music is rhythm. Rhythm refers to the place- piano. Emergence of all untrained relations
ment of sounds in time or the length of time a demonstrated effectiveness of the procedure,
sound is played, and the unit for rhythm is a suggesting that the letters and notes became
beat. The most common rhythms include, but functionally equivalent. Hill et al. (2020) also
are not limited to, whole-notes, half-notes, systematically replicated these results with four
quarter-notes, eighth-notes, sixteenth notes, typically developing children and four children
and dotted half-notes. Each rhythm varies in diagnosed with ASD.
terms of the physical characteristics. For exam- Music educators need to teach numerous
ple, whole notes represent four beats and looks combinations of musical stimuli; thus, matrix
like an open circle, whereas quarter notes repre- training may also function as an effective
sent one beat and looks like a closed circle with method when using auditory–visual MTS.
a vertical line (see Figure 1). During matrix training, participants are taught
When learning to play and interpret music, to respond in the presence of multiple combi-
students must respond similarly to the printed nations of stimuli, then tested on their ability
note (e.g., C), a picture of the note (e.g., see to respond to novel combinations (Pauwels
Figure 1), a dictated name of a note (e.g., “Play et al., 2015). For example, in a 3 x 3 grid of
the note C”), and the sound of the note. In three columns (the colors red, blue, and yellow)
other words, these stimuli must be equivalent. and three rows (the objects boat, car, and
In the behavior-analytic literature, several stud- train), a person is taught to tact some, but not
ies have taught musical relations by establishing all the cells (e.g., red boat, blue car, yellow
equivalence classes among textual, visual, and train). Once these compound stimuli are
auditory stimuli (Arntzen et al., 2010; Griffith learned, the individual components of the pre-
et al., 2018; Hayes et al., 1989; Hill et al., viously trained stimuli are recombined into six
2020; Perez & de Rose, 2010). For example, novel compounds (e.g., red car, blue boat).
Griffith et al. (2018) established three Responding accurately in the presence of these
3-member equivalence classes using auditory– novel compounds without direct teaching has
visual matching-to-sample procedures (MTS) been defined as recombinative generalization
with college students. First, participants were (Goldstein, 1983). Because of the generative
taught to identify the textual representation of learning produced by this teaching structure,
a chord (e.g., C major), select the correct music matrix training seems to be an effective and
notation, and play the chord on the piano efficient tool for teaching a variety of verbal
when presented with the dictated chord name skills (Axe & Sainato, 2010).
(e.g., C major). Researchers tested for the emer- Matrix training may include nonoverlap
gence of playing the chord on the piano in (NOV) or overlap (OV) procedural elements
both the presence of the textual representation (Foss, 1968). When using NOV, the cells
and musical notation of the chord. In addition, (i.e., combinations) that fall along the diagonal
the authors tested to see whether participants of the matrix are taught directly. Each compo-
could identify the correct textual representation nent is taught once in combination with other
in the presence of the music notation components without overlap. For example, if
(i.e., pointing to C chord in the presence of the red boat and blue car are taught in a 2 x
music notation as the sample), as well as the 2 matrix, each component (i.e., blue, red, boat,
symmetrical relation. Participants were also car) is presented only one time (i.e., no over-
tested on whether they could play the previ- lap). In the OV procedure, some components
ously taught relations within a novel sequence are presented more than once during training.
 Matrix Training to Teach Piano Skills 3

Figure 1. Note–rhythm matrix used for the series of experiments. Note. Bold lines separate each submatrix, and each
submatrix was assigned to a specific training procedure. The eleven cells marked ‘Train’ were used for training.

For example, if red boat and red car are taught Investigators have evaluated the use of matrix
in a 2 x 2 matrix, red is presented twice, training when teaching speaker and listener
whereas the objects are presented only once. If behaviors to children with developmental dis-
individuals have no prior knowledge of the abilities across multiple studies (Axe & Sainato,
components, Goldstein (1983) recommended 2010; Frampton et al., 2016; Goldstein &
using OV because it may aid in forming the Brown, 1989; Goldstein & Mousetis, 1989).
key discriminations required for recombinative For instance, Axe and Sainato (2010) taught
generalization. Alternatively, NOV training is four children with autism to engage in different
suggested for learners who already have some actions with writing utensils and pictures
previous knowledge of components in the (i.e., listener responses) using a NOV matrix.
matrix (Foss, 1968; Goldstein, 1983). Three of the four participants demonstrated
4 Emily K. Langton et al.

recombinative generalization of instruction fol- 34 years recruited from a large public univer-
lowing. Frampton et al. (2016) also used the sity. Participation in the study resulted in extra
NOV training structure when teaching children course credit. To ensure that participants had
with autism to tact known nouns (e.g., cat) and no previous experience with music instruction,
verbs (e.g., jumping) together as noun–verb we administered a 10-item questionnaire (see
combinations. Furthermore, Frampton et al. Appendix A), which assessed familiarity with
(2016) probed for generalization across matrices musical terms. Scores below 33% resulted in
with other known nouns and verbs not targeted inclusion in the study. The University’s Institu-
for diagonal training. Four out of five partici- tional Review Board approved all procedures
pants demonstrated recombinative generalization described below.
of tacting noun–verb combinations within the Setting and Materials. Experimental sessions
original matrix, as well as across matrices. were conducted in a 3 x 5 m room, which con-
The main advantage of matrix training is its tained three tables, three computers, and six
overall efficiency. In NOV training, relations are chairs. Sessions ranged from 90 to 120 min
taught directly from the diagonal of the matrix, (M = 110 min) with 5-min breaks offered
and novel untaught combinations may emerge. approximately every 30 min. Sessions took place
For example, if only six of those relations are tau- across 2 days. If 120 min elapsed before a partici-
ght directly within a 6 x 6 matrix, a total of pant completed training, the experimenter sched-
30 new combinations could potentially emerge. uled an additional session within the same week.
Thus, matrix training may potentially take fewer Experimental materials included two three-
trials to teach a set of skills compared with other ring binders (21.59 x 29.54 cm), musical stim-
trial arrangements. When teaching a new skill in uli in clear binder sleeves (21.59 x 29.54 cm), a
which participants have no previous knowledge Casio LK-165 piano keyboard, a Pro Metro-
of either components, an OV matrix training nome android phone application, a music
structure may better lead to a recombinative rep- stand, and a video camera. Musical stimuli
ertoire because it expedites discrimination of the included notes on a music staff (i.e., C, D,
individual components since each component is E, F, G, and A) and rhythms (i.e., whole-note,
paired with another component at least twice half-note, quarter-note, group of eighth-notes,
(Foss, 1968). Although researchers have recently group of sixteenth notes and dotted half-note;
evaluated other training arrangements see Figure 1 for matrix). For all conditions, par-
(i.e., equivalence-based instruction) to teach basic ticipants sat in a chair with the keyboard and
music chords on a piano (Griffith et al., 2018; music stand in front of them. The experi-
Hill et al., 2020), matrix training may also serve menter placed a binder containing the musical
as an efficient tool to teach rhythm and notes. stimuli on the music stand facing participants.
The purpose of this study was to assess the effec- During training conditions, the experimenter
tiveness and efficiency of matrix training in an played a metronome, which emitted short,
OV format to teach college students to tact and steady tones at approximately 60 beats per
play notes and rhythms on the piano. minute. The primary researcher stood approxi-
mately 30 cm to the left of participants, and all
experimental sessions were video recorded. A
EXPERIMENT 1
secondary observer sat adjacent to the primary
Method investigator and behind participants for inter-
Participants. Participants were six undergrad- observer agreement (IOA) and treatment integ-
uate students between the ages of 21 and rity data collection.
 Matrix Training to Teach Piano Skills 5

Experimental Design and Response Measure- vocalizations were inconsistent with the note–
ment. We used a nonconcurrent multiple- rhythm combination (incorrect note, rhythm, or
baseline design across participants (Watson & both) or provided no response within 5 s of the
Workman, 1981) to assess the effects of matrix experimenter’s instruction. During training,
training on the acquisition of tacts and piano experimenters scored tacts as prompted when
playing. Participants were exposed to two sub- participants emitted the correct vocal response
matrices consecutively. Submatrix 1 (i.e., S1) following the experimenter’s vocal prompt.
consisted of nine note–rhythm combinations Experimenters scored play responses as correct
and submatrix 2 (i.e., S2) consisted of 27 note– when participants placed their finger on the cor-
rhythm combinations (see Figure 1). Dependent rect piano key (i.e., played) and depressed the
variables included the percentage of correct key with the correct rhythm in the presence of a
responses during probes and the number of trials given visual or auditory note–rhythm combina-
to reach mastery criterion for tacts and playing tion (e.g., whole-note C) within 5 s. Incorrect
during training (see Table 1). During training, play responses were scored when participants
experimenters collected data on tacts and play played the incorrect piano key or played the
responses to assess if one response component incorrect rhythm or provided no response within
was acquired more quickly than the other using 5 s of the presentation of the note rhythm.
a 5-s response interval for each trial. During Finally, experimenters scored a play response as
music probes and name/play-by-ear probes, prompted when participants played the key with
experimenters collected data on whether partici- the correct rhythm following the experimenter’s
pants could play a sequence of note–rhythm model prompt.
combinations on the piano, and tact and play to Interobserver Agreement and Treatment Integ-
auditory note–rhythm combinations, respec- rity. A second observer assessed IOA either dur-
tively. Experimenters scored tacts as correct ing experimental sessions or via video for an
when participants emitted a vocal response con- average of 99% of sessions for all participants
sistent with either the visual or auditory repre- (range, 97% to 100%). Agreement was defined
sentation of the note–rhythm combination as both the experimenter and second observer
(e.g., saying, “whole-note C”). Experimenters coding the same response per trial (e.g., correct,
scored tacts as incorrect when the participants’ incorrect, or prompted). We used trial-by-trial
agreement to calculate IOA by dividing the
number of agreements by the total number of
Table 1 agreements and disagreements and multiplying
Trials to Criterion by 100. Scores averaged 98% (range, 80% to
Training 1 Remedial 1 Training 2
100%) across all participants and experimental
conditions.
Experiment 1
Neo 140 - 352 A second observer scored treatment integrity
Oracle 180 - 638 by using a checklist that outlined each step of
Trinity 60 - 396
Kali 130 - 176
the procedures for each condition for 99% of
Persephone 180 - 352 sessions (range, 97% to 100%). The second
Morpheus 60 - 242 observer recorded if the step was completed as
Experiment 2
Switch 70 40 506 outlined or if any deviations were observed. We
Ice 160 - 330 calculated treatment integrity scores by dividing
Smith 100 40 154
Niobe 80 50 154 the number of steps completed correctly by the
Cas 190 - 242 total number of steps and multiplying by 100.
Maggie 140 - 330
Scores averaged 99% (range, 80% to 100%)
6 Emily K. Langton et al.

across all participants and experimental During each trial, the experimenter pres-
conditions. ented a page from the binder with a picture of
Procedure. a note–rhythm combination (e.g., C1). Prior to
Pretraining Probe. The purpose of this con- the first block, participants received an instruc-
dition was to measure participants’ tact and tion to repeat after the experimenter in the
play responses to relations to be targeted for presence of a note–rhythm combination and to
training (see relations indicated in Figure 1) to imitate the experimenter after playing a note–
rule out previous history with the experimental rhythm combination on the piano. For exam-
stimuli. During each trial, the experimenter ple, the experimenter delivered a 0-s delay vocal
presented a page in the binder with a picture of prompt (e.g., “This is whole note C) in the
one of the cells from the matrix along with the presence of the given stimulus (e.g., C1). Once
instruction, “What is it?” Once participants the participant repeated the correct response,
responded or 5 s elapsed with no responses, the experimenter provided praise (e.g., “That’s
the experimenter instructed participants to correct”). Next, the experimenter played the
“Play it.” After participants played the piano or note–rhythm combination on the piano and
5 s had elapsed with no response, the experi- provided a statement, such as “This is how you
menter flipped to the next page in the binder play whole note C.” Once the participant imi-
to start the next trial. There were no tated the correct rhythm sequence in the pres-
programmed consequences. The experimenter ence of the note–rhythm combination, the
provided positive statements (e.g., “Keep it experimenter delivered praise.
up.” or “Almost there.”) approximately 2 s after Next, the experimenter implemented a con-
the end of every two trials unrelated to partici- stant 5-s prompt delay for training blocks. For
pants’ performance. Eleven cells, five from S1 example, in the presence of a note–rhythm
and six from S2, (cells labeled “Train”; combination, the experimenter allowed partici-
Figure 1) were presented twice in one 22-trial pants 5 s to independently tact the stimulus
block or three 22-trial blocks for the second upon hearing the instruction, “What is it?”
participant in the dyad. Participants had to Correct tacts were followed by praise
score 33% or below on each 22-trial block. (e.g., “Good!”). Incorrect responses were
Training 1 and 2. During this condition, we followed by an error correction procedure,
taught participants to tact the musical note– which consisted of the experimenter stating the
rhythm combination (e.g., whole-note C) in response was incorrect and vocally prompting
the presence of the visual stimulus and to play the correct response (e.g., “No. It’s whole note
the combination by placing their finger on the C”). If participants did not respond to the vocal
correct piano key in the correct rhythmic prompt, the experimenter would repeat the
sequence (e.g., playing a C note for four prompt one time; however, this did not occur.
counts). In Training 1, we taught five of the Once participants responded to the vocal
cells belonging to S1 (i.e., C1, D1, D2, E2, prompt with the correct tact, the experimenter
and E3; see Figure 1) in a 10-trial block. In delivered the second instruction (e.g., “Play
Training 2, we taught an additional six cells it”). Correct play responses were immediately
belonging to S2 (i.e., F3, F4, G4, G5, A5, and followed by praise, and incorrect responses were
A6) in addition to the S1 cells previously tau- followed by an error correction consisting of
ght for a total of eleven cells in a 22-trial block. the experimenter stating the response was
Training 2 included cells from Training 1 in a incorrect and modeling the correct rhythm
mixed format to maintain responding for the sequence on the piano (e.g., “No. You play it
individual components taught in Training 1. like this”). If participants did not respond to
 Matrix Training to Teach Piano Skills 7

the model, the experimenter would repeat the probed (see Figure 1). During each trial, the
model prompt one time; however, this did not experimenter presented the binder with a pic-
occur. Once participants responded to the ture of one of the four untrained cells from S1
model with the correct rhythm sequence on the or one of the 21 untrained cells from S2. Post-
piano, the experimenter moved to the next training Probe 1 included all S1 untrained cells
trial. Regression criterion to the previous 0-s twice in eight-trial blocks. Posttraining Probe
delay was set at three consecutive trials requir- 2 included all S2 untrained cells twice in
ing error correction. For each trial to be consid- 42-trial blocks. The experimenter only deliv-
ered correct, participants needed to both ered positive statements (e.g., “Keep it up.”)
independently tact and play the note–rhythm approximately 2 s after the end of every two tri-
combination. als regardless of participants’ performance.
Mastery criterion for Training 1 and Train- Passing criterion was set at one block at
ing 2 was set at two blocks at 100% (S1:10 out greater than 88% (7 out of 8) correct
of 10; S2: 22 out of 22) correctly with a metro- responding for Posttraining Probe 1 and greater
nome playing, and one block at 100% correctly than 81% (34 out of 42) correct responding
without the metronome. If the session ended for Posttraining Probe 2. If participants scored
during training (e.g., after 120 min had between 50% and passing, then the experi-
elapsed), the experimenter conducted one menter conducted additional blocks until per-
review block at the start of the next session at a formance reached passing criterion, or after
0-s prompt delay and then progressed to the three blocks of steady responding. If partici-
next delay for the subsequent block. Once par- pants scored below 50% or performance
ticipants met mastery criterion for Training remained steady, then experimenters conducted
1, they progressed to Posttraining Probe 1. Sim- remedial training.
ilarly, once participants met mastery criterion Remedial Trainings 1 and 2. During this
for Training 2, they progressed to Posttraining condition, the experimenter re-trained partici-
Probe 2. pants on the previously taught note–rhythm
Posttraining Probes 1 and 2. The purpose of combinations using the same prompting and
this condition was to test participants’ tact and reinforcement procedures, as well as mastery
play responses in the presence of untrained criterion as in Trainings 1 and 2. Once the
note and rhythm combinations following train- mastery criterion was met, the experimenter
ing. This condition was similar to the Pre- presented the corresponding posttraining
training Probe condition, except the criterion probe.
to progress to the next condition differed Music Probe. The purpose of this condition
slightly. During each probe, the experimenter was to test for generalization of previously tau-
tested responses to the untrained cells ght and untaught cells within the novel context
(e.g., combinations of trained musical compo- of an unfamiliar musical piece (Hill et al.,
nents that were never paired as a combination 2020). Experimenters showed participants up
during the previous training) from either S1 or to three sequences of note–rhythm combina-
S2 depending on which training participants tions in a musical piece and asked them to play
had received prior to this condition the sequence on the piano. The sequence was
(e.g., Training 1 or Training 2). For example, novel to participants; in other words, the exper-
when participants were trained to tact and play imenter did not ask them to play any musical
cells C1 (e.g., whole-note C) and D2 selections that may be familiar to participants
(e.g., half-note D) directly, cells C2 (e.g., half- (e.g., Mary had a little lamb). These musical
note C) and D3 (e.g., quarter-note D) were selections were composed for the purpose of
8 Emily K. Langton et al.

the current study and were novel to the partici- by naming the note–rhythm combination, the
pants. The first block consisted of 19 trials, experimenter delivered a second instruction
while the second block consisted of 21 trials. (e.g., “Play it”) and immediately re-played the
The number of trials differed from the first and sound clip of the note–rhythm combination.
second block to simulate the variation in musi- There were no programmed consequences fol-
cal selections. There were no programmed con- lowing correct or incorrect responses. In addi-
sequences for correct and incorrect playing of tion, no visual stimuli were present
the note–rhythm combinations. The experi- (e.g., binder with music stimuli). Because we
menter only delivered positive statements introduced this condition to measure the possi-
(e.g., “Keep it up.”) approximately 2 s after the ble emergence of auditory to vocal and auditory
end of the music probe regardless of partici- to piano responding, there was no passing
pants’ performance. criterion.
Passing criterion was set at one block at 74% Follow-up Music Probe. The purpose of this
or above (14 out of 19) or one block at 71% condition was to assess whether participants’
or above (15 out of 21). If participants scored performance would maintain over time. We
between 50% and passing, the experimenter conducted a follow-up session (approximately
conducted additional blocks until participants 10 min in duration) 1 week following the last
reached passing criterion, or after three blocks training session for two randomly selected par-
of steady responding. If participants scored ticipants (Trinity and Kali). We only used two
below 50% or performance was steady and participants for the follow-up probe in each
below criterion, then they were exposed to experiment to demonstrate maintenance of
Remedial Trainings 1 and 2. For example, par- music skills over time. We used the same pro-
ticipants would be re-trained on S1 and S2 and cedures and passing criterion described in
the music probe would be re-conducted; how- Music Probe condition.
ever, no participants required remedial training Social Validity. Upon completion of the
following the music probe. study, participants filled out a short survey
Name-/Play-by-Ear Probes 1 and 2. The using a 7-point Likert scale ranging from
purpose of this condition was to test for gener- strongly disagree to strongly agree (Axe &
alized responding in the form of auditory to Sainato, 2010). The scale contained closed-
vocal responding (i.e., name-by-ear) and audi- ended statements about the procedures and
tory to piano playing (i.e., play-by-ear) using outcomes (see Table 2) and provided open-
similar procedures to Hill et al. (2020). Name- ended questions about the participants’ prefer-
by-Ear and Play-by-Ear Probe 1 consisted of ences (e.g., “What did you like about our
the nine S1 relations for total of 18 trials, and teaching methods?” The survey is available
Name-by-Ear and Play-by-Ear probe 2 consisted upon request).
of the 27 S2 relations for a total of 54 trials
(see Figure 1).
In this condition, the experimenter played a Results and Discussion
sound clip of a note–rhythm combination Figures 2 and 3 show the percentage of correct
while the participant was seated in front of the tact and play responses across all probe conditions
piano. During each trial, the experimenter for Neo, Oracle, Trinity, Kali (Figure 2), Per-
stated an instruction (e.g., “What is it?”) and sephone, and Morpheus (Figure 3). All partici-
immediately played the sound clip of the note– pants scored below criterion on the pretraining
rhythm combination belonging to one of the probes. Participants met criterion for Training
submatrices. Once the participant responded 1 after 60 to 180 trials and Training 2 after
 Matrix Training to Teach Piano Skills 9

Table 2 94%). Trinity and Kali were randomly selected

Mean (Range) for Items on Social Validity Survey for to complete Follow-up Music Probe, and Trin-
Experiments 1 and 2 ity passed at 94% correct (18 out of 19) and Kali
Experiment Experiment
passed at 95% correct (20 out of 21). Partici-
Item 1 2 pants completed the experimental conditions in
I learned how to name notes. 6.5 (5-7) 6.7 (6-7) approximately 2-4 hr total ranging from one to
I didn’t learn how to name 1.7 (1-3) 1 (1-3) no more than two sessions for each participant.
notes.a
I learned how to identify 4.7 (4-6) 5 (3-6)
All six participants showed recombinative
rhythms. generalization with piano notes and rhythms
I learned how to play notes. 6.2 (5-7) 6.7 (6-7) following matrix training. These results align
I learned how to play rhythms. 4.8 (3-7) 4.3 (2-6)
I didn’t learn how to play 3.2 (1-6) 2.3 (1-6) with previous research (Axe & Sainato, 2010;
rhythms.a Frampton et al., 2016; Goldstein & Brown,
The music was difficult to learn.a 3.3 (1-6) 3.8 (2-5)
The music was easy to learn. 5 (2-7) 4.7 (4-6) 1989; Goldstein & Mousetis, 1989) in which
I wouldn’t recommend this 2.8 (1-5) 1.3 (1-3) participants demonstrated generative language
teaching strategy for others.a
I feel more prepared to learn 5.5 (4-7) 6.2 (5-7) and recombinative generalization with tacts,
music. and instruction following. To our knowledge,
I do not feel prepared to learn
music.a
2.3 (1-4) 1.2 (1-2)
this procedure was the first successful applica-
I want to learn more about 5.3 (3-7) 5 (4-6) tion of matrix training to music education.
music after the study. Interestingly, we found that while all partici-
I don’t want to learn about 1.7 (1-5) 2.5 (1-4)
music after the study.a pants passed Posttraining Probes 1 and 2, not all
I would recommend this type of 5.3 (3-7) 6 (4-7) of them scored similarly on correct tacts and
music lesson to other people.
I had no formal knowledge of 6 (5-7) 6.2 (5-7) play responses. It is important to note that while
music. there was not an exact correlation between both
I learned to play a song. 4.2 (1-6) 3.7 (2-6)
repertoires (i.e., tacting and playing), percentages
Note: Items were rated on a Likert-type scale of 1 (strongly for both were high, suggesting that tacting the
disagree) to 7 (strongly agree). aItems were reverse coded. note–rhythm combination may have influenced
participants’ subsequent playing of the note–
rhythm combination. For instance, when a par-
176 to 638 trials (see Table 1). Following Train- ticipant incorrectly tacted the note–rhythm com-
ing 1, all participants passed Posttraining Probe bination, the participant was more likely to play
1 with correct play responses ranging from 88% the note–rhythm combination incorrectly as
to 100%, and tact responses ranging from 88% well. However, in some cases, when a partici-
to 100%. Similarly, following Training 2, all pant correctly tacted the note–rhythm combina-
participants passed Posttraining Probe 2 with tion, they made an error on playing and when a
correct play responses ranging from 93% to participant incorrectly tacted the note–rhythm
100%, and tact responses ranging from 93% combination, they actually played the note–
to 100%. rhythm combination correctly (such as is the
Participants scored from 22% to 67% on case for Niobe).
Play-by-Ear Probe 1 and scored from 0% to During Name/Play-by-Ear Probes 1 and
78% on Name-by-Ear Probe 1, which is 2, four of the participants obtained similar
depicted as Tact (A) in figures. For Play-by-Ear scores on correct tacts of the auditory stimulus
Probe 2, participants scored between 16% and and correct play responses following the audi-
44%, and for Name-by-Ear Probe 2, partici- tory stimulus, suggesting that correctly tacting
pants scored from 2% to 41%. Additionally, all the note–rhythm combinations upon hearing a
participants passed Music Probe (range, 71% to sound clip may have aided in correctly playing
10 Emily K. Langton et al.

Figure 2. Results for Neo and Oracle (represented in the left panels); results for Trinity and Kali (represented in the
right panels). Note. The V in parentheses represents visual stimuli and the A in parentheses represents auditory stimuli.

the note–rhythm combination. However, cor- pitch. For example, C and D are closer together
rect tacts of the auditory stimulus and play in pitch, whereas C and A are farther apart.
responses following the auditory stimulus dif- Therefore, C and A are easier to discriminate
fered for two of the participants. For example, than C and D. When hearing the auditory
Trinity scored higher during Play-by-Ear Probe sample of the note C, participants would often
1 (39%) compared to Name-by-Ear Probe tact and play D, whereas when participants
1 (0%). Her errors were due to tacting the note heard the auditory sample of the note A (after
incorrectly; however, she would still play cor- hearing C from a previous trial), they were
rectly in the presence of the auditory compound more likely to correctly tact and play in the
stimulus. Similarly, Morpheus scored higher for presence of the auditory sample.
Play-by-Ear Probe 2 (37%) compared with Consistent with previous music education
Name-by-Ear Probe 2 (2%). His tacts of the research, participants in the current study dem-
auditory stimulus were scored incorrectly onstrated generalized responding by applying
because he only tacted the rhythm of the com- their previously learned piano skills within the
pound stimulus; however, he played the piano context of a musical piece (Arntzen et al.,
accurately in the presence of the auditory com- 2010; Griffith et al., 2018; Hayes et al., 1989;
pound stimulus. It is possible that his correct Hill et al., 2020; Perez & de Rose, 2010). Fur-
play-by-ear responses would have been higher thermore, the two participants who completed
had his tact responses also been correct. a follow-up session maintained their perfor-
Another possible explanation for below profi- mance after 1 week. When interviewed, both
cient Name/Play-by-Ear responding could be participants reported that they had not prac-
participants’ difficulty in discriminating ticed the notes and rhythms or looked up any
between notes that were closer together in music information.
 Matrix Training to Teach Piano Skills 11

Figure 3. Results for Persephone and Morpheus.

One noteworthy limitation is that the metro- tempo. That is, data collectors were determin-
nome present during training was removed dur- ing if the rhythm being played (e.g., whole-
ing probes. Although musicians may practice note) was correct based on their covert account
playing their instrument with a metronome, of tempo, which may have deviated across trials
they typically do not use it during a concert or or sessions, rather than a consistent tempo pro-
a recital. In many cases, musicians may rely on vided by the metronome. In addition, conse-
the gestures made by a conductor, sounds made quences may have been applied inconsistently
by a drummer, or their own behavior (such as as well, depending on the experimenter’s accu-
tapping their feet) to maintain tempo racy of approximately 60 beats per minute.
(i.e., timing). Our removal of the metronome Another potential limitation of Experiment
seemed appropriate for simulating a musician’s 1 was that reinforcement was not faded prior to
natural environment (i.e., concert, recital), but Posttraining Probes 1 and 2. Thus, it is impos-
removing it in an experimental setting increases sible to rule out the possibility that incorrect
the risk of both observer and experimenter drift responses during probes were due to partici-
due to variations in the interpretation of pants’ behavior being placed on extinction.
12 Emily K. Langton et al.

However, given that responding still occurred throughout the entire experiment whereas
during probes, the absence of reinforcement Smith, Niobe, and Maggie had the metronome
likely did not seem to have impacted partici- faded out prior to the posttraining probe condi-
pants’ responding. Finally, during Pretraining tions. Mastery criterion for Switch, Ice, and
Probe, we only assessed for relations that were Cas was set at one 54-trial block at 100% cor-
targeted for training, rather than the new rela- rect responding, and one block at 100% correct
tions tested during Posttraining Probes. This is responding without experimenter feedback
a potential threat to internal validity because (i.e., praise, error correction). Mastery criterion
baseline performance cannot be directly com- for Smith, Niobe, and Maggie was one block at
pared to posttraining performance since differ- 100% correct responding with a metronome
ent relations were tested. We addressed the playing, and one block at 100% correct inde-
aforementioned limitations in Experiment 2. pendent responding without the metronome
and without experimenter feedback.
Pre- and Posttraining Probes 1 and 2. We
EXPERIMENT 2
probed participants’ tacts and play responses to
Method visual stimuli depicting note–rhythm combina-
Participants, Settings, and Materials. tions as in Experiment 1. However, in this
Participants included six typically developing experiment we probed the four note–rhythm
college students between the ages of 22 and combinations from S1 that were not targeted
25 years enrolled at a public university. All par- for training during Training 1. Each of the four
ticipants received extra course credit for com- combinations was presented twice (i.e., 8-trial
pleting the study. block) prior to and after training. We also pro-
Experimental Design and Dependent Vari- bed the 21 note–rhythm combinations from S2
ables. Experimental design and dependent that were not targeted for training during
variables were identical to those described in Training 2. Each of the 21 combinations were
Experiment 1. presented twice (i.e., 42-trial block) prior to
Interobserver Agreement and Procedural Integ- and after training. Mastery criteria were the
rity. Interobserver agreement and treatment same as Experiment 1.
integrity were calculated in the same manner as Results and Discussion. Figure 4 shows the
described in Experiment 1 for 100% of ses- percentage of correct tact and play responses
sions. Interobserver agreement averaged 98% across all probe conditions for Switch, Ice, Cas,
(range, 80% to 100%) and treatment integrity Maggie, Smith, and Niobe. All participants
averaged 99% for all participants (range, 93% scored below criterion on the Pretraining
to 100%). Probe. Participants met criterion for Training
Procedures. All procedures were conducted as 1 after 70 to 190 trials and Training 2 after
described in Experiment 1 apart from Pre- and 154 to 506 trials (see Table 1). Following
Posttraining Probes 1 and 2 (described below). Training 1, Ice, Cas, and Maggie passed Post-
Three of the participants (Switch, Ice, and Cas) training Probe 1 with correct play responses
were exposed to the metronome during all of the ranging from 88% to 100%, and with correct
conditions, while the other three participants tacts ranging from 88% to 100%; however,
(Smith, Niobe, and Maggie) were exposed to the Switch, Niobe, and Smith failed the Post-
metronome during training only, as in Experi- training Probe 1 with correct play responses
ment 1. ranging from 0% to 50% and with correct tacts
Trainings 1 and 2. During training, Switch, also ranging from 0% to 50%. Following 40 to
Ice, and Cas had the metronome playing 50 trials of Remedial Training 1 (i.e., Training
 Matrix Training to Teach Piano Skills 13

Figure 4. Results for Switch, Ice and Cas (represented in the left panels); results for Smith, Niobe, and Maggie (rep-
resented in the right panels)

1), Switch scored 100% on the Posttraining During the name/play-by-ear probes, partici-
Probe 1 for both play and tact responses on the pants scored between 17% to 78% on Play-by-
third attempt and Smith and Niobe scored Ear Probe 1 and 0% to 67% on Name-by-Ear
88% on the first attempt for both play and tact Probe 1. Participants scored between 20% and
responses. Following Posttraining Probe 1, all 59% on Play-by-Ear Probe 2 and between 0%
participants were exposed to a Pretraining and 44% on Name-by-Ear Probe 2. Subse-
Probe 2, in which all participants scored below quently, all participants passed Music Probe
criterion. Following Training 2, all participants (range, 73% to 95%). Smith and Cas were ran-
passed Posttraining Probe 2 with play responses domly selected to return for Follow-up Music
ranging from 81% to 95%, and with correct Probe, and both participants passed (range,
tacts ranging from 88% to 100%. Participants 74% to 95%).
completed the experimental conditions in The purpose of Experiment 2 was to address
approximately 4-5 hr total ranging from two to the limitations of Experiment 1 by examining
no more than three sessions for each the effects of a) keeping the metronome con-
participant. stant across all conditions versus removing the
14 Emily K. Langton et al.

metronome at the end of each training condi- Training 2 is the fact that notes are represented
tion, b) probing untrained relations prior to by letters of the alphabet. The notes progress
training, and c) removing reinforcement prior on a music staff in the same manner as letters
to posttraining probe conditions. Potential progress in the alphabet. For example, notes
observer and experimenter drift were ruled out progress in the following sequence: A, B, C,
by having the metronome present for Switch, D, E, F, G, A and so on. Since notes C, D,
Ice, and Cas, while the metronome was and E were taught during Training 1, it is pos-
removed for Smith, Niobe, and Maggie at the sible that Cas learned that F was the next note
end of each training. Results showed that the in the sequence, based on her previous learning
presence of the metronome did not affect par- history with the alphabet.
ticipants’ performances during probes. Probing As a result of matrix training, all participants
untrained relations prior to and after training passed Posttraining Probe 2 following Training
ensured that measurement of the dependent 2 with no need for Remedial Training 2. These
variable was directly related to the training results show that participants who needed
itself. In addition, removing reinforcement Remedial Training 1 (Switch, Smith, Niobe)
prior to posttraining probe conditions did not did not need Remedial Training 2. In addition,
differentially affect performance. Overall, the all participants maintained their performance
results were similar to those obtained in Experi- when tested on Posttraining Probe 1 relations
ment 1. after Training 2.
Three out of six participants required reme- As in Experiment 1, participants demon-
dial training after Training 1, which may have strated generalization by playing piano notes
been due to participants attending to only one and rhythms in the form of a musical piece
component of the compound stimulus. For (Griffith et al., 2018; Hill et al., 2020). Smith
example, they may have attended to the physi- and Cas were invited to come back 1 week fol-
cal characteristics of the note (e.g., open, lowing their last training to complete Follow-
closed) and not the placement of the note on up Music Probe. Smith and Cas both passed
the music staff. After Remedial Training 1, no Follow-up Music Probe.
additional trainings were required. Also, all par-
ticipants demonstrated minimal correct
responding during Name/Play-by-Ear Probes GENERAL DISCUSSION
1 and 2. Matrix training was effective in teaching col-
None of the participants met passing crite- lege students to play piano note–rhythm com-
rion for Pretraining Probe 2 tact and play after binations that were never presented together
Training 1, but Cas demonstrated some, albeit during training, replicating results from previ-
minimal, emergence of correct responding. ous research (Axe & Sainato, 2010; Foss, 1968;
During Pretraining Probe 2, Cas sometimes Frampton et al., 2016; Goldstein et al., 1987;
tacted and played correctly in the presence of Goldstein & Brown, 1989; Goldstein &
the visual representation of recombined notes Mousetis, 1989). Across both experiments, par-
and rhythms. She correctly tacted and played ticipants completed Training 1 in 60 to 190 tri-
in the presence of the visual representation of als and Training 2 in 154 to 638 trials.
the note F when combined with previously Training 2 likely took longer for participants to
trained rhythm components (e.g., whole-note F, master because 11 relations were targeted for
half-note F) which had not been directly tau- training from the matrix (i.e., S2), whereas
ght. One possible explanation for Cas’s perfor- Training 1 only included five relations from
mance during Pretraining Probe 2 prior to the matrix (i.e., S1). It is possible that the
 Matrix Training to Teach Piano Skills 15

relations taught in Training 2 may have the note after it was presented. This behavior
required fewer trials to reach mastery if taught may have helped Niobe attend to and select the
together rather than separating the matrix into correct key on the piano. The experimenters
two training conditions. did not observe other participants engage in
Overall, participants rated the procedures any other mediating behaviors like Niobe; how-
and outcomes favorably (see Table 2). Eleven ever, it is possible they were covertly rehearsing
out of the 12 participants stated they would the pitch.
learn music with matrix training again. Partici- Morpheus, Niobe, and Cas showed the most
pants stated they would have preferred more pronounced differences between percentage of
feedback throughout training in terms of how correct play responses and percentage of correct
to utilize the metronome and how to place tact responses during the Name/Play-by-Ear
their hands on the piano correctly. probes. Playing and tacting may be two func-
During Experiment 2, three participants tionally independent repertoires. Playing may
(Switch, Smith, and Niobe) did not demon- involve conditional stimulus control, given that
strate recombinative generalization with stimuli when hearing the sound (conditional stimulus),
presented in novel combinations. One common the player must scan the keyboard, see the cor-
error among these three participants was that rect key (discriminative stimulus), press it, and
they consistently tacted incorrectly in the pres- keep it pressed for the correct amount of time.
ence of the visual representation of the note On the other hand, tacting involves simple dis-
C. The note C has a line through it , criminative control by the compound note–
rhythm combination over a vocal topography.
whereas the other notes either fall immediately
Interestingly, for these two participants, the
below, on a line, or in between lines within the
auditory compound stimuli exerted stronger
staff, which may have not been salient to par-
discriminative control over playing the piano
ticipants. Thus, these results suggest specific during Play-by-Ear Probes 1 and 2 than tacting
components lacked discriminative control over following auditory stimuli during Name-By-Ear
participants’ responses when those components Probes 1 and 2 (i.e., speaker).
were paired with other components in novel Persephone responded correctly during
combinations (Miguel et al., 2015). It is also Name/Play-by-Ear 2 in 9 of 54 trials (17%).
possible that participants’ errors may have been We hypothesize that with this participant, the
due to limited exposure to the note C. During response products of her tacting may have
Training 1, the note C is only presented once served as SDs for playing during those trials;
as a trained relation (e.g., whole-note C), however, this was not consistent across partici-
whereas notes D and E are presented twice pants. For example, correctly tacting in the
(e.g., whole-note D, half-note D, half-note E, presence of the auditory compound stimulus
quarter-note E). Participants may have needed did not help improve Morpheus’, Niobe’s, and
additional learning opportunities to attend to Cas’ playing performance. These participants
compounds consisting of note C. played more than they tacted correctly during
The purpose of Name/Play-by-Ear Probes Name/Play-by-Ear Probe 1. Thus, our results
1 and 2 was to assess for the possible emer- suggest that the product of the participants’
gence of tacting and playing in the presence of tacting did not always serve as an SD for
sound clips following matrix training alone. playing the note (Miguel, 2016; 2018). How-
Interestingly, some participants almost scored ever, it is possible that if participants had been
to proficiency during these probes. In Niobe’s directly trained to play note–rhythm combina-
case, she would occasionally hum the pitch of tions in the presence of their respective
16 Emily K. Langton et al.

auditory samples (a form of listener training) Although performances did not differ, each
prior to learning to tact note–rhythm combina- group included only three participants. Future
tions (speaker training), then the auditory studies may need to further evaluate the role of
product of their tact would have served as an this variable with additional participants.
SD for playing (Ribeiro et al., 2015). Future Fourth, we did not require participants to
studies should assess the effects of listener play at a specific tempo when performing in
matrix training (i.e., playing the piano in the the presence of a visual representation of a
presence of auditory samples) on tacting and musical piece during the music probe. Partici-
playing in the presence of compound stimuli pants could take as much time as they needed
(e.g., Lee et al., 2015). to play. Playing with timing is an important
There are some additional limitations and component of music and would add to the
future directions worth mentioning. First, par- external validity of the current procedures. We
ticipants had to tact the note–rhythm combina- also did not teach participants to respond cor-
tion correctly before playing it on the piano rectly in the presence of visual rests. Rests are
key, which entailed pressing the correct key music symbols that represent the absence of
and holding it for a specific number of counts. playing. When performing, rests are incorpo-
It is possible that tacting the note–rhythm rated into a musical piece and musicians should
combination may have facilitated pressing the be counting covertly when they see rests on
correct piano key and playing the correct their sheet music. Future research should
rhythm. It is unclear whether the same results explore how to teach rests by counting out loud
would have been obtained if participants had or taking a breath. In addition, the length of
been required to play the correct note–rhythm the session and small number of sessions
combination prior to tacting it. Future required for mastery may lessen the generality
researchers may want to investigate whether of the current treatment procedures for adults
changing the order of tacting and playing music with disabilities or typically-developing children
notes would yield different results. due to the practicality and likelihood of these
Second, the notes on the music staff progress populations practicing in this manner.
in the same order as the alphabet, which corre- Finally, we taught only six notes and six
spond to the location of the piano key. For rhythms, which is not sufficient to teach some-
instance, C, D, E, F, G, and A are on the one the necessary skills to become a proficient
piano keys in the exact same order as they pro- pianist. Future researchers can systematically
gress on the music staff. Thus, it is possible replicate these procedures by expanding the size
that participants quickly learned to tact each of the matrix in terms of number of notes and
note accurately and find the correct piano key rhythms or by using three-dimensional matrices
due to their prior knowledge of the alphabet. (Goldstein & Mousetis, 1989). Additionally, it
The speed at which individuals in this study would be important to compare the efficiency
learned to tact and play piano notes and of the matrix training implemented in this
rhythms may not be replicated when teaching study to other methods (e.g., NOV matrix
typically developing children or children with training, stimulus equivalence).
disabilities, especially if they do not have prior Music may be considered a form of verbal
knowledge of the alphabet. behavior. Playing music may fit Skinner’s
Third, in Experiment 2, half of the partici- (1957) definition because when learning music,
pants were exposed to the metronome across all the teacher (i.e., listener) has been trained spe-
conditions, while the other half was only cifically to respond to the behavior of the
exposed to the metronome during training. speaker (i.e., the player). Reynolds and Hayes
 Matrix Training to Teach Piano Skills 17

(2017) argue that Skinner’s verbal operants linguistic repertoires of language learners. Analysis and
could be analogous to musical behaviors. For Intervention in Developmental Disabilities, 3, 279–293
https://doi.org/10.1016/0270-4684(83)90002-2.
example, echoics may be analogous to mimicry Goldstein, H., & Brown, W. H. (1989). Observational
or copying sounds on their instrument. Even learning of receptive and expressive language by
though a student may practice his or her music handicapped preschool children. Education and Treat-
ment of Children, 12, 5–37. https://www.jstor.org/
alone, a history of socially mediated conse- stable/42899091.
quences is responsible for the initial develop- Goldstein, H., & Mousetis, L. (1989). Generalized lan-
ment of the musical repertoire, which can now guage learning by children with severe mental retar-
dation: Effects of peers’ expressive modeling. Journal
be maintained as the speaker (player) serves as of Applied Behavior Analysis, 22, 245–259 https://doi.
his or her own listener (Horne & Lowe, 1996; org/10.1901/jaba.1989.22-245.
Miguel, 2016). If music is in fact verbal, then Gourgey, C. (1998). Music therapy in the treatment of
the same procedures used to teach other forms social isolation in visually impaired children. Rehabili-
tation and Education for Blindness and Visual Impair-
of verbal responding could be employed to ment, 29, 157–162.
advance research in music education (Greer, Graziano, A. B., Peterson, M., & Shaw, G. (1999).
1980; Reynolds & Hayes, 2017). Enhanced learning of proportional math through
music training and spatial-temporal training. Neuro-
logical Research, 21, 138–152 https://doi.org/10.
1080/01616412.1999.11740910.
REFERENCES Greer, R. D. (1980). Design for music learning. Teachers
College Press.
Arntzen, E., Halstadtro, L., Bjerke, E., & Halstadtro, M.
Griffith, K. R., Ramos, A. L., Hill, K. E., &
(2010). Training and testing music skills in a boy
Miguel, C. F. (2018). Using equivalence-based
with autism using a matching-to-sample format.
instruction to teach piano skills to college students.
Behavioral Interventions, 25, 129–143 https://doi.org/
Journal of Applied Behavior Analysis, 51, 207–219
10.1002/bin.301.
https://doi.org/10.1002/jaba.438.
Axe, J. B., & Sainato, D. M. (2010). Matrix training of
Hanna-Pladdy, B., & MacKay, A. (2011). The relation
preliteracy skills with preschoolers with autism. Jour-
between instrumental musical activity and cognitive
nal of Applied Behavior Analysis, 43, 635–652 https://
aging. Neuropsychology, 25, 378–386 https://doi.org/
doi.org/10.1901/jaba.2010.43-635.
10.1037/a0021895.
Brandt, A., Gebrian, M., & Slevc, L. R. (2012). Music Hayes, L. J., Thompson, S., & Hayes, S. C. (1989). Stim-
and early language acquisition. Frontiers in Psychology, ulus equivalence and rule following. Journal of the
3, 327 https://doi.org/10.3389/fpsyg.2012.00327. Experimental Analysis of Behavior, 52, 275–291
Brown, L. S., & Jellison, J. A. (2012). Music research https://doi.org/10.1901/jeab.1989.52-275.
with children and youth with disabilities and typically Hill, K. E., Griffith, K. R., & Miguel, C. F. (2020).
developing peers: A systematic review. Journal of Using equivalence-based instruction to teach piano
Music Therapy, 49, 335–364 https://doi.org/10. skills to children. Journal of Applied Behavior Analysis,
1093/jmt/49.3.335. 53(1), 188–208 https://doi.org/10.1002/jaba.547.
Farber, A., & Parker, L. (1987). Discovering music Horne, P. J., & Lowe, C. F. (1996). Reflections on nam-
through Dalcroze Eurhythmics. Music Educators Jour- ing and other symbolic behavior. Journal of the Exper-
nal, 74, 43–45 https://doi.org/10.2307/3397940. imental Analysis of Behavior, 65, 315–340 https://doi.
Foss, D. J. (1968). An analysis of learning in a miniature org/10.1901/jeab.1996.65-315.
linguistic system. Journal of Experimental Psychology, Hudziak, J. J., Albaugh, M. D., Ducharme, S.,
76, 450–459 https://doi.org/10.1037/h0025506. Karama, S., Spottswood, M., Crehan, E., …
Frampton, S. E., Wymer, S. C., Hansen, B., & Botteron, K. N. (2014). Cortical thickness matura-
Shillingsburg, M. A. (2016). The use of matrix train- tion and duration of music training: health-
ing to promote generative language with children promoting activities shape brain development. Journal
with autism. Journal of Applied Behavior Analysis, 49, of the American Academy of Child & Adolescent Psychi-
869–883 https://doi.org/10.1002/jaba.340. atry, 53, 1153–1161 https://doi.org/10.1016/j.jaac.
Gilbert, J. P. (1983). A comparison of the motor music 2014.06.015.
skills of nonhandicapped and learning disabled chil- Kaviani, H., Mirbaha, H., Pournaseh, M., & Sagan, O.
dren. Journal of Research in Music Education, 31, (2014). Can music lessons increase the performance
147–155. https://doi.org/10.2307/3345217. of preschool children in IQ tests? Cognitive Processing,
Goldstein, H. (1983). Recombinative generalization: Rela- 15, 77–84 https://doi.org/10.1007/s10339-013-
tionships between environmental conditions and the 0574-0.
18 Emily K. Langton et al.

Kim, J., Wigram, T., & Gold, C. (2008). The effects of Rapp, J. (2007). Further evaluation of methods to identify
improvisational music therapy on joint attention behav- matched stimulation. Journal of Applied Behavior Analysis,
iors in autistic children: A randomized controlled study. 40, 73–88 https://doi.org/10.1901/jaba.2007.142-05.
Journal of Autism and Developmental Disorders, 38, Reynolds, B. S., & Hayes, L. (2017). Parallels and incon-
1758–1766 https://doi.org/10.1007/s10803-008-0566-6. gruities between musical and verbal behaviors. The
Lee, G. P., Miguel, C. F., Darcey, E. K., & Psychological Record, 67, 413–421 https://doi.org/10.
Jennings, A. M. (2015). A further evaluation of the 1007/s40732-017-0221-8.
effects of listener training on the emergence of Ribeiro, D. M., Miguel, C. F., & Goyos, C. (2015). The
speaker behavior and categorization in children with effects of listener training on discriminative control
autism. Research in Autism Spectrum Disorders, 19, by elements of compound stimuli in children with
72–81 https://doi.org/10.1016/j.rasd.2015.04.007. disabilities. Analysis of Verbal Behavior, 104, 48–62
Lim, H. A. (2010). Use of music in the applied behavior https://doi.org/10.1002/jeab.161.
analysis verbal behavior approach for children with Russell-Smith, G. (1967). Introducing Kodaly principles
autism spectrum disorders. Music Therapy Perspectives, into elementary teaching. Music Educators Journal,
28, 95–105 https://doi.org/10.1093/mtp/28.2.95. 54, 43–46. https://www.jstor.org/stable/3391185.
Lim, H. A., & Draper, E. (2011). The effects of music Schellenberg, E. (2006). Long-term positive associations
therapy incorporated with applied behavior analysis between music lessons and IQ. Journal of Educational
verbal behavior approach for children with autism Psychology, 98, 457–468 https://doi.org/10.1037/
spectrum disorders. Journal of Music Therapy, 48, 0022-0663.98.2.457.
532–550 https://doi.org/10.1093/jmt/48.4.532. Schonbrun, M. (2005). The everything reading music book:
Mannell, R. C. (2007). Leisure, health, and well-being. A step-by-step instruction to understanding music nota-
World Leisure Journal, 49, 114–128 https://doi.org/ tion and theory. Adams Media.
10.1080/04419057.2007.9674499. Skinner, B. F. (1957). Verbal behavior. Appleton-Cen-
McRae, S. W. (1982). The Orff connection…reaching tury-Crofts.
the special child. Music Educator’s Journal, 68, 32–34 Sussman, J. (2009). The effect of music on peer awareness
https://doi.org/10.2307/3395960. in preschool age children with developmental disabil-
Miguel, C. F. (2016). Common and intraverbal bidirec- ities. Journal of Music Therapy, 46, 53–68 https://doi.
tional naming. The Analysis of Verbal Behavior, 32, org/10.1093/jmt/46.1.53.
125–138 https://doi.org/10.1007/s40616-016-0066-2. Suzuki Association of the Americas, Inc. (2011). What is
the Suzuki method? http://suzukiassociation.org.
Miguel, C. F. (2018). Problem-solving, bidirectional nam-
ing, and the development of verbal repertoires. Behav- Vaughn, K. (2000). Music and mathematics: Modest sup-
ior Analysis Research and Practice, 18, 340–353 port for the oft-claimed relationship. Journal of Aes-
https://doi.org/10.1037/bar0000110. thetic Education, 34, 149–166. https://www.jstor.org/
stable/3333641.
Miguel, C. F., Frampton, S. E., Lantaya, C. A.,
Watson, P. J., & Workman, E. A. (1981). The non-
LaFrance, D. L., Quah, K., Meyer, C. S., …
concurrent multiple baseline across-individuals
Fernand, J. K. (2015). The effects of tact training on
design: An extension of the traditional multiple base-
the development of analogical reasoning. Journal of
line design. Journal of Behavior Therapy and Experi-
the Experimental Analysis of Behavior, 104, 96–118
mental Psychiatry, 12, 257–259 https://doi.org/10.
https://doi.org/10.1002/jeab.167.
1016/0005-7916(81)90055-0.
Pauwels, A. A., Ahearn, W. H., & Cohen, S. J. (2015). Zuk, J., Andrade, P. E., Andrade, O. A., Gardiner, M., &
Recombinative generalization of tacts through matrix Gaab, N. (2013). Musical, language, and reading abili-
training with individuals with autism spectrum disor- ties in early Portuguese readers. Frontiers in Psychology,
der. The Analysis of Verbal Behavior, 31, 200–214 4, 288. https://doi.org/10.3389/fpsyg.2013.00288.
https://doi.org/10.1007/s40616-015-0038-y.
Perez, W. F., & de Rose, J. C. (2010). Recombinative
generalization: An exploratory study in musical read- Received February 7, 2019
ing. Analysis of Verbal Behavior, 26, 51–55 https:// Final acceptance November 30, 2019
doi.org/10.1007/BF03393082. Action Editor, Alice Shillingsburg
 Matrix Training to Teach Piano Skills 19

APPENDIX A

Music Questionnaire
Please answer the following questions the best you can. Ask the experimenter if you have any
questions.
1. What rhythm is this? _________________________

2. Circle the note ‘c’.

3. What note is this? _________________________

4. Draw a quarter note. _________________________

5. What notes are these?

_____ _____ _____ _____ _____

6. Draw the note ‘d’ on the staff.
7. What is this note? _________________________

8. What rythmns are these?

_____ _____ _____ _____ _____
9. What is this rythmn? _________________________

10. Circle the half note.